It's tough to come in cold with an answer to this question as there are a lot of variables involved, such as how X3 was installed on your 50 user site. Maybe MongoDB and SQL And Syracuse are all installed on the same machine, maybe they are separate, not sure where to point you to first without some basic such as what services exist on which machines (or maybe it's a mono-tier install). 

But, if you don't have those details either and you are going in cold to a project like this, one thing you can do, since you know which machine has the problem is to put a tool like WinDirStat on the machine (or any other tool you prefer that does something like this). That will at least tell you what files are taking the most space and WinDirStat at least does a good job helping you determine if it is one big file or many small files in a group that are causing your outage. And, if you know the file structure that is bloating over time you can at least have a good idea as to which service is causing this bloat. 

Thanks for your inputs.

In the mean time we discover there is 2.55 approx adonix_memory_*.tra files in E:\Sage\X3RUN\runtime\tmp. Each file is 24Kb : (24 * 1024) * 2,550,000 = ~60GB

There is also *.tra files in E:\Sage\X3V12\folders\KGHPRD\TRA (2.7GB just for yesterday). 

In the E:\Sage\X3RUN\runtime\tmp since 12 April 2022 until yesterday morning there was 483 files, since 10:28 am yesterday until now there are the remainder of the 2.5 million files.

Something changes seriously somewhere

Any chance you have a file in your runtime\tmp directory named start_log?

If so, remove it. That will cause logging by the runtime for as long as it exists. 

Hi Braam Koekekemoer,
Do you witness a lot of session crashing perhaps ?
2.55 million looks very odd though.

Regarding TRA and tmp, this is a normal housekeeping to do every month.

check your SOAP/REST web services and batch tasks as well.

Bruno

We found the issue. It seems there are two web app concepts for mobility in Sage X3. Once of them using the TCWPSHTOSIGM representation with class TCWPSH. The other one use the PSHTOSIGM representation with class TCAWRKHISSUI. The last one mentioned seems going into a loop the moment any action in the mobile is done e.g. approve, rejects etc, and it bombards the server with 25k *.tra files literally hundred per minutes until the server runs out of space.

wow... Good catch!
I believe you need Sage Support to be involved as you might have found a bug.
Did you try raising maxheap value ? Maxheap is the equivalent of maxmem but for classes. It might be crashing because it lacks available memory.